Method for acquiring texture of 3D model and related apparatus

Abstract

A method for acquiring a texture of a three-dimensional (3D) model includes: acquiring at least two 3D networks generated by a target object based on a plurality of angles, the at least two 3D networks including a first correspondence between point cloud information and color information of the target object, and first camera poses of the target object; acquiring an offset between 3D points used for recording the same position of the target object in the at least two 3D networks according to the first camera poses respectively included in the at least two 3D networks; updating the first correspondence according to the offset, to acquire a second correspondence between the point cloud information and the color information of the target object; and acquiring a surface color texture of a 3D model of the target object according to the second correspondence.

What is claimed is: 
     
       1. A method for acquiring a texture of a three-dimensional (3D) model, performed by a computing device, the method comprising:
 acquiring at least two 3D networks generated based on a target object according to a plurality of angles, the at least two 3D networks including a first correspondence between point cloud information and color information of the target object, and first camera poses of the target object, the first camera pose being used for representing a displacement of the target object relative to a reference position in response to determining that the 3D network is generated; 
 moving the at least two 3D networks to a same angle according to the first camera poses respectively included in the at least two 3D networks; 
 acquiring a second point closest to a first point in a first network, the second point being in a second network, and the first network and the second network being respectively different 3D networks of the at least two 3D networks; 
 acquiring an offset between the first point and the second point; 
 updating the first correspondence according to the offset, to acquire a second correspondence between the point cloud information and the color information of the target object; and 
 acquiring a surface color texture of a 3D model of the target object according to the second correspondence. 
 
     
     
       2. The method according to  claim 1 , wherein acquiring the second point closest to the first point in the first network comprises:
 traversing points in the second network; 
 respectively acquiring 3D coordinates of the points in the second network; 
 respectively calculating distances between the points in the second network and the first point according to the 3D coordinates; and 
 determining a point closest in the second network to the first point as the second point. 
 
     
     
       3. The method according to  claim 1 , wherein acquiring the second point closest to the first point in the first network comprises:
 acquiring the second point closest to the first point by using a k-Nearest Neighbor (kNN) algorithm. 
 
     
     
       4. The method according to  claim 1 , wherein updating the first correspondence comprises:
 acquiring a pixel in the color information corresponding to a 3D point in the point cloud information according to the first correspondence; 
 offsetting the 3D point in the point cloud information according to the offset; and 
 acquiring a correspondence between the 3D point acquired after being offset in the point cloud information and the pixel as the second correspondence. 
 
     
     
       5. The method according to  claim 4 , wherein the offset includes a rotation matrix R used for representing a rotation operation and a translation matrix T used for representing a translation operation, and offsetting the 3D point in the point cloud information according to the offset comprises:
 executing the following formula: D1=(R|T)×D2, 
 D1 being point cloud information in a 3D network, and D2 being information in another 3D network. 
 
     
     
       6. The method according to  claim 4 , wherein acquiring the surface color texture of the 3D model of the target object comprises:
 acquiring the pixels respectively corresponding to the 3D points of the 3D model according to the second correspondence; and 
 covering the pixels on the corresponding 3D points to implement texture mapping on a surface of the 3D model. 
 
     
     
       7. The method according to  claim 1 , wherein acquiring the at least two 3D networks comprises:
 acquiring at least two initial images of the target object at a plurality of shooting angles, the at least two initial images respectively recording depth information of the target object, the depth information being used for recording a distance between each point of the target object and the reference position, and the reference position being a position of a camera that photographs the target object; 
 performing back-projection in a 3D space according to the depth information in each initial image, to acquire first point cloud information corresponding to each initial image, points in the first point cloud information being used for recording the 3D points of the target object; 
 acquiring the first correspondence between the 3D point and the pixel in the color information; and 
 generating the 3D network according to the first point cloud information and the first correspondence. 
 
     
     
       8. An apparatus for acquiring a texture of a three-dimensional (3D) model, comprising: a memory storing computer program instructions; and a processor coupled to the memory and configured to execute the computer program instructions and perform:
 acquiring at least two 3D networks generated based on a target object according to a plurality of angles, the at least two 3D networks including a first correspondence between point cloud information and color information of the target object, and first camera poses of the target object, the first camera pose being used for representing a displacement of the target object relative to a reference position in response to determining that the 3D network is generated; 
 moving the at least two 3D networks to a same angle according to the first camera poses respectively included in the at least two 3D networks; 
 acquiring a second point closest to a first point in a first network, the second point being in a second network, and the first network and the second network being respectively different 3D networks of the at least two 3D networks; 
 acquiring an offset between the first point and the second point; 
 updating the first correspondence according to the offset, to acquire a second correspondence between the point cloud information and the color information of the target object; and 
 acquiring a surface color texture of a 3D model of the target object according to the second correspondence. 
 
     
     
       9. The apparatus according to  claim 8 , wherein the processor is further configured to execute the computer program instructions and perform:
 traversing points in the second network; 
 respectively acquire 3D coordinates of the points in the second network; 
 respectively calculate distances between the points in the second network and the first point according to the 3D coordinates; and 
 determine a point in the second network closest to the first point as the second point. 
 
     
     
       10. The apparatus according to  claim 8 , wherein the processor is further configured to execute the computer program instructions and perform:
 acquiring the second point closest to the first point by using a k-Nearest Neighbor (kNN) algorithm. 
 
     
     
       11. The apparatus according to  claim 8 , wherein the processor is further configured to execute the computer program instructions and perform:
 acquiring a pixel in the color information corresponding to a 3D point in the point cloud information according to the first correspondence; 
 offsetting the 3D point in the point cloud information according to the offset; and 
 acquiring a correspondence between the 3D point acquired after being offset in the point cloud information and the pixel as the second correspondence. 
 
     
     
       12. The apparatus according to  claim 11 , wherein the offset includes a rotation matrix R used for representing a rotation operation and a translation matrix T used for representing a translation operation, and offsetting the 3D point in the point cloud information according to the offset includes:
 executing the following formula: D1=(R|T)×D2, 
 D1 being point cloud information in a 3D network, and D2 being information in another 3D network. 
 
     
     
       13. The apparatus according to  claim 11 , wherein acquiring the surface color texture of the 3D model of the target object includes:
 acquiring the pixels respectively corresponding to the 3D points of the 3D model according to the second correspondence; and 
 covering the pixels on the corresponding 3D points to implement texture mapping on a surface of the 3D model. 
 
     
     
       14. The apparatus according to  claim 8 , wherein acquiring the at least two 3D networks includes:
 acquiring at least two initial images of the target object at a plurality of shooting angles, the at least two initial images respectively recording depth information of the target object, the depth information being used for recording a distance between each point of the target object and the reference position, and the reference position being a position of a camera that photographs the target object; 
 performing back-projection in a 3D space according to the depth information in each initial image, to acquire first point cloud information corresponding to each initial image, points in the first point cloud information being used for recording the 3D points of the target object; 
 acquiring the first correspondence between the 3D point and the pixel in the color information; and 
 generating the 3D network according to the first point cloud information and the first correspondence. 
 
     
     
       15. A non-transitory computer-readable storage medium storing computer program instructions executable by at least one processor to perform:
 acquiring at least two 3D networks generated based on a target object according to a plurality of angles, the at least two 3D networks including a first correspondence between point cloud information and color information of the target object, and first camera poses of the target object, the first camera pose being used for representing a displacement of the target object relative to a reference position in response to determining that the 3D network is generated; 
 moving the at least two 3D networks to a same angle according to the first camera poses respectively included in the at least two 3D networks; 
 acquiring a second point closest to a first point in a first network, the second point being in a second network, and the first network and the second network being respectively different 3D networks of the at least two 3D networks; 
 acquiring an offset between the first point and the second point; 
 updating the first correspondence according to the offset, to acquire a second correspondence between the point cloud information and the color information of the target object; and 
 acquiring a surface color texture of a 3D model of the target object according to the second correspondence. 
 
     
     
       16. The non-transitory computer-readable storage medium according to  claim 15 , wherein acquiring the second point closest to the first point in the first network includes:
 traversing points in the second network; 
 respectively acquiring 3D coordinates of the points in the second network; 
 respectively calculating distances between the points in the second network and the first point according to the 3D coordinates; and 
 determining a point closest in the second network to the first point as the second point. 
 
     
     
       17. The non-transitory computer-readable storage medium according to  claim 15 , wherein acquiring the second point closest to the first point in the first network includes:
 acquiring the second point closest to the first point by using a k-Nearest Neighbor (kNN) algorithm. 
 
     
     
       18. The non-transitory computer-readable storage medium according to  claim 15 , wherein updating the first correspondence includes:
 acquiring a pixel in the color information corresponding to a 3D point in the point cloud information according to the first correspondence; 
 offsetting the 3D point in the point cloud information according to the offset; and 
 acquiring a correspondence between the 3D point acquired after being offset in the point cloud information and the pixel as the second correspondence. 
 
     
     
       19. The non-transitory computer-readable storage medium according to  claim 18 , wherein the offset includes a rotation matrix R used for representing a rotation operation and a translation matrix T used for representing a translation operation, and offsetting the 3D point in the point cloud information according to the offset comprises:
 executing the following formula: D1=(R|T)×D2, 
 D1 being point cloud information in a 3D network, and D2 being information in another 3D network. 
 
     
     
       20. The non-transitory computer-readable storage medium according to  claim 19 , wherein acquiring the surface color texture of the 3D model of the target object includes:
 acquiring the pixels respectively corresponding to the 3D points of the 3D model according to the second correspondence; and 
 covering the pixels on the corresponding 3D points to implement texture mapping on a surface of the 3D model.